What's changed: Added per-section figures (cert-figure-retrofit). New AI-200 Chapter 1 (vector data = embeddings/distance metric/ANN index/semantic vs keyword search, Azure vector-enabled data services = Azure AI Search/Cosmos DB for NoSQL/PostgreSQL+pgvector/Azure Managed Redis and selection criteria)
1.2Azure Vector-Enabled Data Services
Understand Azure options for storing/searching AI app vectors—Azure AI Search, Cosmos DB for NoSQL, Azure Database for PostgreSQL (pgvector), and Azure Managed Redis—and how to choose by use case, from a developer’s view.
Vectors can be stored/searched not only in dedicated "vector DBs" but by adding vector capability to existing data services. AI-200 tests choosing the right service by use case, existing assets, and operations. Know the main options.
- Azure AI Search: search-specialized; offers hybrid (vector + keyword) and semantic ranking—a go-to RAG retrieval store.
- Azure Cosmos DB for NoSQL: globally distributed, low-latency NoSQL with vector search, co-locating app data and vectors.
- Azure Database for PostgreSQL (pgvector): add vector columns to a relational DB via the pgvector extension—leverage SQL and existing data.
- Azure Managed Redis: in-memory, ultra-low latency—good for vector search and semantic caching.
1.2.1Selection criteria
Decide by "closeness to existing data," "latency needs," and "search sophistication." Need advanced search (hybrid/ranking)? Azure AI Search. Want to co-operate with app NoSQL data? Cosmos DB for NoSQL. Leverage relational assets? PostgreSQL + pgvector. Need ultra-low latency or caching? Azure Managed Redis. There is no single right answer—choose by requirements.
Common: (1) "RAG retrieval store with advanced hybrid search/ranking" = Azure AI Search. (2) "vector search co-located with app NoSQL data" = Cosmos DB for NoSQL. (3) "add vector columns to an existing relational DB" = PostgreSQL + pgvector. (4) "ultra-low latency / semantic cache" = Azure Managed Redis. Choose by requirements.
Watch out: (1) these are not exclusive—choose by use, and combine. (2) pgvector is a PostgreSQL extension, not a separate service. (3) "need a vector DB" does not mean a dedicated product—existing services can gain vector capability. (4) Advanced search (hybrid/ranker) is Azure AI Search’s strength.
1.2.2Section summary
- Vectors can be stored/searched by adding vector capability to existing services, not only dedicated DBs
- Azure AI Search = advanced search / Cosmos DB for NoSQL = co-locate app data / PostgreSQL+pgvector = relational assets / Azure Managed Redis = ultra-low latency
- Not exclusive—choose by requirements (existing data/latency/search sophistication)
Sign in to track progress — Log in.
Quick check
(just a quick review)Q1. Which Azure service offers hybrid (vector+keyword) search and semantic ranking as a go-to RAG retrieval store?
Q2. To add vector columns to an existing relational DB for vector search, which fits?
Q3. For globally distributed, low-latency NoSQL with vector search co-located with app data, which fits?
Q4. Which in-memory, ultra-low-latency Azure service suits vector search and semantic caching?
Q5. Which is correct about choosing Azure vector-enabled data services?